Octozi Raises $3M to Clean Up Clinical Trial Data With Agentic AI

A controlled study of Octozi's platform found that reviewer error rates on clinical trial data dropped from 54.7% to just 8.5%, a result the New York-based startup is now backing with fresh capital. Octozi has closed a $3 million seed round led by Surface Ventures, with participation from Debiopharm's venture arm, which had previously backed the company.

Octozi builds agentic AI tools for data cleaning and query management in clinical development, a part of the drug trial pipeline that remains notoriously manual and error-prone. Sponsors and contract research organizations routinely employ large teams of reviewers to catch inconsistencies in patient records, lab results, and case report forms before a trial can move toward regulatory submission. Octozi's pitch is that much of this work can be automated without sacrificing the auditability regulators expect.

The platform pairs large language models with deterministic clinical algorithms, an architecture meant to combine the pattern-recognition strengths of LLMs with the predictability required in a regulated environment. Rather than replacing human reviewers outright, Octozi is built around a human-in-the-loop workflow, routing flagged issues to clinical staff while automating the repetitive triage that currently consumes much of their time.

The numbers behind the seed round are notable. Beyond cutting reviewer error rates by more than 80%, Octozi's internal study reported roughly a six-fold increase in data cleaning throughput and a fifteen-fold reduction in false positive queries, the incorrect flags that send clinical teams chasing issues that don't actually exist. Octozi's own economic modeling translates those gains into more than $5 million in projected savings per Phase III oncology trial, a figure that speaks directly to the cost pressures sponsors face as trials grow larger and more data-intensive.

"We built Octozi to give clinical teams their time back without asking them to trust a black box," said Amit Patel, the company's co-founder and CEO, framing the platform's deterministic-plus-LLM design as a middle path between full automation and the status quo of manual review.

Gyan Kapur, managing partner at Surface Ventures, is backing the bet that clinical operations, long underserved by modern software, represent fertile ground for agentic AI specifically because the tasks are repetitive, rules-based, and high-stakes enough that errors carry real cost. Debiopharm's continued involvement, following its earlier investment, suggests the pharmaceutical side of the industry sees similar potential in narrowing the gap between AI capability and the compliance demands of clinical trials.

With the new funding, Octozi is positioned to expand beyond its initial deployments and test whether its error-reduction numbers hold up across a broader range of trial types and sponsors, a question that will determine whether agentic AI can meaningfully dent the cost and timeline of clinical development at scale.